Would Accountability Based on Teacher Value Added Be Smart Policy? An Examination of the Statistical Properties and Policy Alternatives

With annual standardized student testing, it now may be feasible to measure the contributions to student achievement made by individual teachers. But will these "teacher value- added" measures help to improve student achievement and learning? I address this question through a "policy validity" framework" that includes three factors: (a) statistical validity, i.e., how well teacher value-added measures actually measure true teacher contributions to achievement; (b) purposes, i.e., whether the measures are intended to signal effective teachers or provide a path to improvement; and (c) costs. Regarding statistical properties, I outline many of the key assumptions of value-added, most of which are rejected by empirical evidence. However, there is some evidence that the assumption violations may not be severe and that the measures contain useful information. I also compare the policy validity of teacher value-added accountability to three main policy alternatives: teacher credentials, school value-added accountability, and formative uses of test data. This analysis shows that teacher value-added is likely to increase student achievement more effectively and efficiently than a teacher credentials-only strategy, but it is unclear whether teacher value-added would raise achievement more than alternative uses of student test scores. Resolving this issue will require studies of how different uses of student test scores influence instruction and student learning in practice.